Coding Agents
10 pieces on Coding Agents.
News & Analysis
JetBrains Mellum2.1: 2.5B Active Params Reach 47 SWE-bench Verified
JetBrains' 12B MoE model activates 2.5B parameters per token and lifts SWE-bench Verified from 2.0 to 47.0 after reinforcement learning in real repositories.
SWE-2 Matches Fable 5.1 on FrontierCode at 64% Lower Cost
Cognition's SWE-2 scores 50.0% on FrontierCode 1.1 Main, within 1 point of Fable 5.1, while costing 64% less — post-trained via RL on Kimi K3's 2.8T parameters.
OpenAI Logs 3.1 Agent-Workdays Per Human Workday by August 2026
OpenAI's transparency report shows median researchers spending $600+/day on inference, 3.1 agent-workdays per human workday, and Astra GPU cuts of 59.2%.
Meta Pays ~95% Discount for Muse Spark Users Who Share Agent Data
Meta's Muse Spark slashes API prices by up to 95% for users who share prompts and outputs — putting an explicit dollar figure on agentic trace data.
Qwen 3.8 27B Is Capable but Defaults to Extreme Overthinking
Alibaba's Apache 2.0-licensed 27B vision model fits in 17 GB and handles agents, vision, and code — but its xhigh reasoning default is a trap.
Qwen 3.8 27B Is Strong but Overthinks by Default
Alibaba's 17 GB Qwen 3.8 27B excels at vision, tool use, and coding agents — but its xhigh reasoning default burns tokens on trivial prompts.
Z.ai GLM-5.3: Post-Training Gains on a Fixed 743B Base Model
GLM-5.3 reuses GLM-5.2's 743B base unchanged. Terminal-Bench 3.0 jumps from 4.6 to 28.3; CyberGym hits 84.5%, edging closed frontier models.
Codex Comes to the ChatGPT Mobile App in Preview
Codex is moving onto phones, letting developers start, steer, approve, and monitor AI work from the ChatGPT app.
DeepSWE Reshuffles the AI Coding Leaderboard and Puts GPT-5.5 on Top
A tougher coding benchmark shows wider gaps between frontier AI models, with GPT-5.5 leading and verifier quality becoming the real story.
OpenClaw Creator's $1.3M OpenAI Token Bill Shows the New Cost of Agentic Coding
A reported $1.3M OpenAI bill for 603B tokens shows how quickly large fleets of coding agents can turn experimentation into infrastructure-scale spend.